The Alpine Intonation of Bern Swiss German

THE ALPINE INTONATION OF BERN SWISS GERMAN

Jennifer Fitzpatrick-Cole University of Konstanz, Germany

ABSTRACT Phonological aspects of intonation can be categorized according to the phonological form of tunes, the inventory of This paper presents data from the Bern dialect of Swiss German phonological tunes, and the meanings assigned to phonological from a typological perspective. Bern Swiss German is tunes. impressionistically more ÒalpineÓ than Northern Standard Phonological form differences include phonotactics, tonal German, partly as a result of their distinct Òdefault accentsÓ, shapes, (e.g. monotonal vs. bitonal pitch accents or phrasal H*+L in NSG and L*+H BSG. Similar dialectal variation has tones), well-formedness conditions on tunes (e.g. OCP effects), also been reported for English. Like Northern Standard German, distribution of tunes, tune-text association, phonological rules Bern Swiss German can associate L phrase tones to the end of (e.g. deletion, Tone Linking, possibly truncation), phonological focus or to a postnuclear stressed syllable. Assuming a phrasing, deaccenting, and so forth. typological framework which recognizes distinctions between Systemic differences refer to the intonational distinctions phonetic and phonological aspects of intonation as well as languages make, that is, what set of tunes each language has in its universal and language-specific aspects, the differences between inventory. For instance, Glasgow English neutralizes the BSG and NSG discussed here are taken to be phonological rather distinction between statements and questions ([14, 16]), which than phonetic. other dialects maintain with distinct contours. English and German neutralize neutral focus with narrow focus on the 1. TYPOLOGY OF INTONATION rightmost element, while Bengali distinguishes them with two Work on intonation is slowly gaining ground in typology (see for different contours. instance [13]). Ladd [14] shows that BolingerÕs [2, 3, 4] Semantic differences refer to the meaning and use of tunes. ÒuniversalistÓ theory of intonation is steadily losing ground to As soon as we adopt the phonological/typological view of ÒphonologicalÓ theories of intonation (beginning with Bruce [5] intonation, we practically take it for granted that languages and and Pierrehumbert [17]). The former contends that intonation is dialects use different tunes for the same meaning or have extralinguistic, essentially emotional. The closest it comes to the different meanings for the same tune. grammar is the phonetics, meaning that any typological There are still numerous aspects of intonation that are not generalizations to be made about it are not like those made for included in this table, such as discourse factors and disfluencies. the grammar. In the latter, intonation has a decidedly phonological component distinct from the phonetic implementation, and thereby cross-linguistic differences can be either phonetic or phonological. A sketch for a typology of intonation is given in Table 1.1 This is slightly modified from Ladd [14], who adapts WellsÕ [20] taxonomy of segmental variation to intonational variation, and incorporates GussenhovenÕs [10] distinction between the phonetic and phonological levels of intonation. The first cut is between the extralinguistic Ñ which includes some of the aspects of intonation that Bolinger considered truly universal, including communicative functions like the imparting of emotion or interest Ñ and the linguistic. Under linguistic, the next cut is between the universal and the language-specific. BolingerÕs universals are vague tendencies that all have counterexamples, but they can be salvaged with ÒweakÓ versions. Under language-specific, the next cut is between the phonetic and the phonological. The traditional distinction is that gradient differences are phonetic and categorical differences are phonological. As for phonetics, we can separate the universal aspects like intrinsic pitch of segments from the language- specific, phonetic-y, non-meaningful aspects, such as late vs. early peak alignment, interpolation vs. spreading of static tones, possibly compression vs. truncation, and perhaps various downtrends and uptrends, such as downstep/catathesis, upstep, downdrift, and final lowering.

page 941 ICPhS99 San Francisco

Extra- linguistic

Òstrong universalist viewÓ [2, 3, 4] Òweak universalist viewÓ [14]

(i) Completion, finality, declaratives: (i«) Declaratives do not necessarily fall from the low/falling pitch (= LI) last pitch accent ¥ Glasgow English Q/decl. [16] L*H HP LI ¥ East Bengali some declaratives LI HI

(ii) Incompleteness, non-finality, questions: (ii«) Questions have high/rising pitch near the end high/rising pitch (= HI) ¥ Hungarian Q [14] L* HP LI Universal ¥ Bengali yes/no Q [11] L* HI LI

(iii) New/salient information: (iii«) New/Salient information is marked with a local pitch peaks (= T*) pitch accent and possibly a ¥ subordinate postnuclear pitch accent Romanian Q [14] L* H* LI ¥ a lexical pitch accent Bengali focus clitics [15] L* H* ¥ a phrasal tone Bengali focus [11] L* HP

Phonetic differences Languages can have different phonetic realizations of the same phonological tune, e.g. early vs. late peak alignment, downstep/catathesis, upstep, Linguistic downdrift, final lowering, interpolation vs. spreading.

Phonological differences Systemic differences Languages can have different phonological forms Languages can have different inventories of of tunes, e.g. phonotactics, tonal shapes, well- distinct phonological tunes. formedness conditions on tunes, distribution of tunes, tune-text association, phonological rules, ¥ Standard English declarative H* LI phonological phrasing, deaccenting. yes/no question L* HI Language ¥ Glasgow English decl., ques. L*H HP LI specific ¥ Bengali monotonal boundary tones HI LI ¥ Bengali focus declarative L* HP LI bitonal boundary tones HILI LIHI neutral declarative H* LI ¥ German, English ¥ German, English declarative H*+L LI monotonal boundary tones HI LI

¥ German, Eng. Òhat patternÓ H H Semantic differences ¥ Bengali no plateaus Languages can differ in the meaning or use of phonological tunes.

¥ English, German focus H*+L ¥ Bengali neutral focus H* LI

¥ Bern German L*+H ¥ Standard N. German H*+L

Table 1: Typology of Intonation

page 942 ICPhS99 San Francisco

2. BERN SWISS GERMAN 2.1. Default pitch accent 2.2. Focus Whereas the default pitch accent in NSG (1) is typically H*+L, The accent patterns in simple SOV sentences are what we would in BSG it is typically L*+H. expect from what we know of German and Dutch ([1, 12, 10, 6]). An example of a sentence with no narrow focus is given in (3). (1) Northern Standard German The F0 peak is often realized following the stressed syllable, 250 there is a fairly gradual fall to the next stressed syllable, and a H*+L !H*+L LI typical downstep pattern. 200 (3) (Answer to ÔWhat happened?Õ) 350 150 L*+H L*+!H LI 300

100 250

200 5 0 Fahnder deu---tet auf einen Be-su---cher-----stuhl. 150 ÔThe detective pointed to a visitorÕs chair.Õ E Fa-n‡----ti--ker het e Po--l’---ti--ker er-m—r------det. 100 A fanatic has a politician murdered. 0.5 1.01.5 5 0 1.0 2.0 3.0 (2) Bern Swiss German 250 Examples with narrow focus are given in (4Ð6). (4) shows L*+H L*+!H LI narrow focus on the subject and ÒdeaccentingÓ of the object. 200 There is a clear L target at the end of the focus, which remains flat until the final fall (see ¤2.3). Uhmann [21] analyzes a similar post-focus L in German as the trail tone of H*+L, which spreads 150 rightwards to the next focus domain boundary or the next word stress. The analogous analysis for (4) might be L*+H+L, but I 100 have opted for a phrase tone LP (LÐ) to mark the end of focus.

5 0 Dr Fang----er dŸ---tet uf e Bsue---cher-----stuehl. (4) (Answer to ÔWho murdered the politician?Õ) ÔThe detective pointed to a visitorÕs chair.Õ 350 L*+H LP LI 300 0.5 1.01.5 250

The default accent in BSG has all the hallmarks of a rising 200 accent L*+H. L* aligns to the stressed syllable, is often a step 150 down from a preceding syllable, is low for a good portion of the stressed syllable, and is a rise from low pitch into a peak that 100 [E Fana-----ti--ker]F het dr Po-li--ti--ker er-mor-det. occurs much later than ToBI L+H* [18]. The +H peak can occur 50 after the stressed syllable (even if the stressed syllable is long, 1.0 2.03.0 perhaps unlike ToBI). The fall usually begins right after the In (5) the lack of downstep on the object, cf. (3), gives the peak, before the end of the word; indicating +H rather than HP (HÐ), which would sustain the H to the end of the word or phrase. narrow focus reading, which can be interpreted as being the The fall has two sources: in prenuclear position, L*+H is object or the VP. In German and English objects and verbs followed by another L*+H; in nuclear position, L*+H is followed phrase together, and a focus accent on the object can take scope over the VP (cf. Bengali). The fact that the object in (5) is not a phrase tone LP (see ¤2.2). Under this analysis the difference between BSG and NSG is not a realizational difference at the immediately followed by a L suggests that the object and verb phonetic level, but a semantic difference at the phonological phrase together, and the post-focus L in (4) and (5) does indeed level. In work on English dialects, Mayo et al. [16] and Grabe align to the end of the focused phrase (where it meets the and Nolan [8] have analyzed some English dialects in the same sentence final LI in (5)). fashion, where some have H*+L and others have L*+H. BSGÕs default accent could also be phonologically (5) (Answer to ÔWho did the fanatic murder?Õ or ÔWhat did the fanatic do?Õ) represented as H*+L, but phonetically implemented with a (very) 350 late F0 peak alignment. Such an analysis, however, pushes our L*+H L*+H LP LI typological model of intonation to the point of being useless if 300 our only interest is making an (unnecessary) link to NSG. A 250 plausible case of phonetic peak-timing difference is proposed by Grabe [7] for Northern Standard German and Southern British 200 English. Phonologically they have the same default accent 150 Dr Fa-na-----ti----ker het [e Po--li--ti--ker]F er-mor----det. H*+L, but differ on the phonetic level in terms of the peak 100 alignment: NSG aligns the peak later in the stressed syllable than Dr Fa-na-----ti----ker het [e Po--li--ti--ker er-mor----det]F. 50 SBE (and both are later than Romance). 2.0 3.0

page 943 ICPhS99 San Francisco

ACKNOWLEDGMENTS Narrow focus on the verb is also indicated by a pitch accent This research was supported by a grant from the Deutsche L*+!H (6). Here the object also has L*+!H, and is Forschungsgemeinschaft Schwerpunktprogramm ÒSprachtypologieÓ unambiguously in the background, i.e. not focused. The subject, NOTES like all previous examples, has L*+H, and may or may not be 1. In the notation used here, T = ToBI [18] T%, an Intonation Phrase interpreted as focused. I boundary tone and Tp = ToBI TÐ, a Phonological/Intermediate Phrase boundary tone. (6) (Answer to ÔWhat happened to the politician?Õ or ÔWhat did the fanatic do to the politician?Õ) REFERENCES 350 [1] Bierwisch, M. 1966. Regeln fŸr die Intonation deutscher SŠtze. Studia L*+H L*+!H L*+!H LI 300 Grammatica, 7, 99-201. [2] Bolinger, D. L. 1978. Intonation across languages. In Greenberg, J. 250 (ed.), Universals of Human Language. Stanford: Stanford University 200 Press. 471-524. [3] Bolinger, D. L. 1986. Intonation and its Parts. Palo Alto: Stanford 150 University Press. [E Fa-na--ti-ker]F het dr Po-li-ti--ker [er-mor--det.] F [4] Bolinger, D. L. 1989. Intonation and its Uses. Palo Alto: Stanford 100 E Fa-na---ti-ker het dr Po-li-ti--ker [er-mor--det.] F University Press. 50 [5] Bruce, G. 1977. Swedish Word Accents in Sentence Perspective. 1.0 2.0 Lund: Gleerup. [6] FŽry, C. 1993. German Intonational Patterns. Vol. 285. TŸbingen: 2.3. Final L Niemeyer. In (3Ð5) there is not a gradual fall at the end of the utterance but a [7] Grabe, E. 1998. Comparative Intonational Phonology: English and sharp fall on the stressed syllable of the verb, suggesting that German. Ph.D. dissertation, University of Nijmegen. there is a final L that associates to the final stressed syllable [8] Grabe, E., Nolan, F., and Farrar, K. J. 1998. The phonetic and rather than the final boundary. phonological representation of cross-varietal differences in English One analysis is to revise the analysis of the default pitch intonation. Poster, Labphon6, York. accent as H*+L and allow +L to associate to the following [9] Grice, M. and BenzmŸller, R. 1998. Tonal affiliation in German stressed syllable, even if it is unaccented (as a variation of Partial nuclear falls and fall-rises. Poster, Labphon6, York. [10] Gussenhoven, C. 1984. On the Grammar and Semantics of Sentence Tone Linking). This would account for the L on the verb and Accents. Dordrecht: Foris. perhaps the apparent L* (incorrectly) assumed earlier. Curiously, [11] Hayes, B. and Lahiri, A. 1991. Bengali intonational phonology. however, this +L can only associate to the last stress in the Natural Language and Linguistic Theory, 9, 47Ð96. phrase, as in (4), where it can associate to the verb but not the [12] Isacenko, A. v. and SchŠdlich, H. J. 1966. Untersuchungen Ÿber die object. More problematic, (4) has two L targets following the deutsche Satzintonation. Studia Grammatica 7. Berlin: Akademie-Verlag. pitch accent, so +L probably does not account for both of them. [13] Hirst, D., and Di Cristo, A. (eds.) 1998. Intonation Systems: A Another analysis is to transcribe H+L* or L* on the verb. In Survey of Twenty Languages. Cambridge: Cambridge University Press. an analysis of southern German dialects, Truckenbrodt [19] [14] Ladd, D. R. 1996. Intonational Phonology. Cambridge: Cambridge transcribes SOV sentences with no narrow focus as L*+H L*+H University Press. [15] Lahiri, A. and Fitzpatrick-Cole, J. 1999. Emphatic clitics and focus H+L*. In (4,5), however, there is a preceding narrow focus on intonation in Bengali, in Kager, R. and Zonneveld, W. (eds.), Phrasal the subject or object, so the final H+L* on the verb would not be Phonology. Dordrecht: Foris. 119Ð144. the nuclear pitch accent. In Standard German verbs in such [16] Mayo, C., Aylett, M., and Ladd, D. R. 1997. Prosodic transcription constructions only receive pitch accents under narrow focus, of Glasgow English: an evaluation study of GlaToBI. In Botinis, A., which corresponds to (6), where the verb has L*+H. Moreover, a Kouroupetroglou, G, and Carayiannis, G. (eds.), Proceedings of the post-focus object receives no pitch accent, as in (4), so this ESCA Tutorial and Research Workshop on Intonation: Theory, Models potential post-focus pitch accent on the verb in (4,5) would be and Applications. 231Ð234. special to verbs. [17] Pierrehumbert, J. 1980. The Phonology and Phonetics of English Grice and BenzmŸller [9] show that the L target in German Intonation. Ph.D. dissertation, MIT. [18] Silverman, K., Beckman, M., Pitrelli, J., Ostendorf, M., Wightman, nuclear falls and fall-rises also falls on a postnuclear stressed C., Price, P., Pierrehumbert, J. and Hirschberg, J. 1992. ToBI: A standard syllable. They argue that treating L as part of the preceding for labeling English prosody. International Conference on Spoken nuclear pitch accent, H*+L, would predict that L be a fixed Language Processing. distance from the H peak. Instead, they suggest the L is [19] Truckenbrodt, H. 1998. On the Role of Prosody in Tonal Scaling. independent of the pitch accent, similar to LaddÕs [14] analysis of Talk presented at the Workshop on Focus, Intonation and Phrasing, certain postnuclear tones in English, Greek, and Romanian. A Freudental, Germany, 3/19/98. postnuclear phrase tone is independent of, but subordinate to, the [20] Wells, J. C. 1982. Accents of English. Cambridge: Cambridge nuclear pitch accent. It is accent-like in some languages in that it University Press. associates to a stressed syllable if one is available, otherwise it [21] Uhmann, S. 1991. On the tonal disambiguation of focus structures. Journal of Semantics 8, 219Ð238. behaves like a boundary tone. Along these lines, I suggest that the final L in (3Ð5) is the Intonation Phrase boundary tone LI (L%), which has the option of being stress-seeking.